Price, R. W., T. Dang and J. Botev (2015), “Adjusting fiscal balances for the business cycle: New tax and expenditure elasticity estimates for OECD countries”, OECD Economics Department Working Papers, No. 1275, OECD Publishing, Paris.
http://dx.doi.org/10.1787/5jrp1g3282d7-en
OECD Economics Department Working Papers No. 1275
Adjusting fiscal balances for the business cycle
NEW TAX AND EXPENDITURE ELASTICITY ESTIMATES FOR OECD COUNTRIES
Robert W. Price, Thai-Thanh Dang, Jarmila Botev
JEL Classification: E62, H30, H60
Organisation de Coopération et de Développement Économiques
Organisation for Economic Co-operation and Development 10-Dec-2015
___________________________________________________________________________________________
_____________ English - Or. English
ECONOMICS DEPARTMENT
ADJUSTING FISCAL BALANCES FOR THE BUSINESS CYCLE: NEW TAX AND EXPENDITURE ELASTICITY ESTIMATES FOR OECD COUNTRIES
ECONOMICS DEPARTMENT WORKING PAPERS No. 1275
By Robert W.R. Price, Thai-Thanh Dang and Jarmila Botev
OECD Working Papers should not be reported as representing the official views of the OECD or of its member countries. The opinions expressed and arguments employed are those of the author(s).
Authorised for publication by Jean-Luc Schneider, Deputy Director, Policy Studies Branch, Economics Department.
All Economics Department Working Papers are available at www.oecd.org/eco/workingpapers
JT03388129
Complete document available on OLIS in its original format
This document and any map included herein are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.
ECO/WKP(2015)93Unclassified English - Or. Eng
OECD Working Papers should not be reported as representing the official views of the OECD or of its member countries. The opinions expressed and arguments employed are those of the author(s).
Working Papers describe preliminary results or research in progress by the author(s) and are published to stimulate discussion on a broad range of issues on which the OECD works.
Comments on Working Papers are welcomed, and may be sent to the Economics Department, OECD, 2 rue André-Pascal, 75775 Paris Cedex 16, France, or by e-mail to [email protected].
All Economics Department Working Papers are available at www.oecd.org/eco/workingpapers.
The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities.
The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law.
© OECD (2015)
You can copy, download or print OECD content for your own use, and you can include excerpts from OECD publications, databases and multimedia products in your own documents, presentations, blogs, websites and teaching materials, provided that suitable acknowledgment of OECD as source and copyright owner is given. All requests for commercial use and translation rights should be submitted to [email protected]
ABSTRACT/RÉSUMÉ
Adjusting fiscal balances for the business cycle:
new tax and expenditure elasticity estimates for OECD countries
This paper re-estimates the elasticities of government revenue and expenditure items with respect to the output gap for OECD countries. These elasticities are used by the OECD to calculate cyclically adjusted fiscal balances. The study updates the earlier 2005 study using the most recent datasets and tax codes, the coverage being confined in this paper to 35 countries, the 34 OECD member states and Latvia.
The same two-step methodology is retained: revenue and expenditure elasticities with respect to the output gap being defined as the product of, first, the elasticities of individual revenue and expenditure items with respect to their bases and, second, the elasticities of these bases with respect to the output gap. A number of refinements and methodological improvements are made relative to the 2005 study. The revisions to individual elasticities relative to the 2005 vintage are significant in a number of cases but do not follow a clear pattern across countries, except for the elasticities of corporate income tax revenue which are revised up in most cases.
JEL classification codes: E62, H30, H60.
Key words: budget elasticity, automatic stabilisers, fiscal surveillance, cyclically adjusted.
********************
Correction des soldes budgétaires en fonction des variations cycliques :
nouvelles estimations d’élasticités des impôts et des dépenses pour les pays de l’OCDE
Cet article estime les élasticités des composantes de revenus et de dépenses des administrations publiques par rapport aux écarts de production pour les pays de l’OCDE. Ces élasticités sont utilisées par l’OCDE pour calculer les soldes financiers des administrations publiques corrigés du cycle économique.
Cette étude est une mise à jour des travaux parus en 2005, elle utilise les données et les codes d’impôts les plus récentes , et couvre 35 pays, à savoir les 34 pays membres ainsi que la Lettonie. La méthode en deux étapes a été conservée : les élasticités par rapport aux écarts de production étant définies comme le produit , dans un premier temps, des élasticités des composantes individuelles de recettes et de dépenses par rapport à leurs assiettes , et dans un deuxième temps des élasticités de ces assiettes par rapport aux écarts de production. Des modifications et des améliorations méthodologiques ont été apportées depuis l’étude de 2005. Les révisions d’élasticités par rapport à la version de 2005 sont importantes dans certains cas mais ne suivent pas un schéma type pour tous les pays, à l’exception des élasticités des impôts sur les bénéfices des sociétés qui ont été révisées à la hausse dans la plupart des cas.
Classification JEL : E62, H30, H60.
Mots clefs : élasticité budgétaire, stabilisateurs automatiques, surveillance fiscale, ajustement cyclique.
TABLE OF CONTENTS
ADJUSTING FISCAL BALANCES FOR THE BUSINESS CYCLE: NEW TAX AND EXPENDITURE
ELASTICITY ESTIMATES FOR OECD COUNTRIES ... 6
I. Introduction ... 6
II. Conceptual and methodological issues ... 7
III. Revenue elasticities ... 11
IV. Government expenditure elasticities ... 23
V. Aggregate budget sensitivity and cyclically adjusted budget estimates ... 26
BIBLIOGRAPHY ... 36
METHODOLOGICAL AND STATISTICAL ANNEX ... 38
I. COMPUTATION OF INCOME-TAX AND SOCIAL SECURITY-CONTRIBUTION ELASTICITIES ... 38
II. ESTIMATING INDIRECT AND CORPORATE TAX ELASTICITIES ... 45
III. ESTIMATING TAX BASE/OUTPUT GAP ELASTICITIES ... 50
Tables Table 1. Categorisation of taxes and transfers and their bases ... 10
Table 2. Tax to tax base elasticities of income and social security contributions. ... 13
Table 3. Source of revisions of tax to tax base elasticities for personal income and social security contributions ... 14
Table 4. Computation of tax to tax base elasticity of personal income and its components ... 16
Table 5. Tax revenue to output gap elasticities of personal income and social security contributions ... 17
Table 6. Corporate tax elasticities with respect to profits and the output gap ... 20
Table 7. Indirect tax elasticities with respect to the output gap ... 22
Table 8. Government expenditure elasticities ... 25
Table 9. Summary of revenue elasticities with respect to the output gap ... 27
Table 10. Budget balance ratio semi-elasticities ... 29
Table 11. Actual and cyclically-adjusted budget balances using new elasticity estimates ... 30
Table A.1.1 Derivation of weighted average elasticity from individual tax codes ... 39
Table A1.2. Personal income component tax and output gap elasticities ... 44
Table A1.3. Corporate tax to gross operating surplus regressions ... 47
Table A1.4. VAT/GST tax to private consumption regressions ... 48
Table A1.5. Non-tax revenue elasticities ... 49
Table A1.6. Wage and salaries to output gap estimations... 52
Table A1.7. Self-employment income to output gap regressions ... 53
Table A1.8. Corporate income tax base to output gap regressions ... 54
Table A1.9. Capital income to output gap regressions... 55
Table A1.10. Adjusted elasticities for wages and salaries, self-employment incomes and gross operating surplus with respect to the output gap ... 56
Table A1.11. Unemployment to output elasticity regressions ... 57
Table A1.12. Revisions of personal income and social security tax to tax base elasticities ... 58
Figures
Figure 1. Actual and cyclically-adjusted balances using 2005 and new elasticity estimates ... 32
Figure A1.1. Marginal and average tax rates with income distribution ... 38
Figure A1.2. Tax elasticity effects of changes in allowances and rates ... 40
Figure A1.3. Effects of income distribution revisions on tax elasticities ... 40
Figure A1.4. Cyclical adjustment under an ECT model ... 46
Boxes Box A1. Cyclical adjustment in the ECT model ... 45
ADJUSTING FISCAL BALANCES FOR THE BUSINESS CYCLE: NEW TAX AND EXPENDITURE ELASTICITY ESTIMATES FOR OECD COUNTRIES
By Robert W.R. Price, Thai-Thanh Dang and Jarmila Botev1
I. Introduction
1. This paper presents new cyclically adjusted budget balance estimates for OECD member economies, based on revised and updated estimates of tax and expenditure elasticities. The existing elasticities date back to 2005, and relate to 2003 tax codes and tax base information (referred to as the 2005 model throughout this paper) (Girouard and André, 2005). The sample period used for all the regressions is 1990-2013.2 The new elasticities incorporate information from the latest available tax and benefit codes (as of 2013, for most countries) and more up-to-date tax base information. The coverage is also extended to include the new OECD member states.3,4 The paper is a sequel to the study on EU country elasticities published in December 2014 and used to update the European Commission’s cyclical adjustment process (Mourre et al., 2014)5, but is based on the output gap methodology used in the context of the OECD's forecasting exercise which does not coincide exactly with that of the European Commission (EC). As such, the results for EU economies will not exactly match those used by the Commission, but will essentially be very close.
2. The study broadly applies the same method as the 2005 one, which used tax code information to derive revenue to base elasticities and econometric analysis to derive the relationship between bases and the cycle. However, while the approach is broadly the same as in Girouard and André (2005), the paper introduces some refinements to the methodology in order to ensure that the cyclical adjustment model corresponds more closely to reality. The revisions are the following:
In addition to making use of more informative and up-to-date income-distribution data, the calculation of aggregate personal income tax elasticities separately identifies the major tax base components of personal income: earned income, self-employment income and capital income.
Previously, only taxes on wages were considered.
1. Robert W.R. Price is a consultant economist, Thai-Thanh Dang is a consultant econometrician and Jarmila Botev is an economist with the OECD Economics Department. Corresponding author:
[email protected]. The authors would like to thank Christophe André, Caterina Astarita, Sven Blondal, Yvan Guillemette, Annabelle Mourougane, Gilles Mourre, Thomas Neubig, Savina Princen, members of the Output Gap Working Group of the EU’s Economic Policy Committee and delegates of the OECD Working Group on Short-Term Economic Prospects for comments and suggestions on earlier drafts; Sylvie Foucher-Hantala for statistical help; and Isabelle Fakih for help with document preparation.
2. For individual countries, the sample period may be shorter, due to data availability. For more detailed information on data sources and sample periods used for calculation of each elasticity, see the Appendix.
3. Latvia, which is in the process of accession, is also included.
4. OECD averages presented in the paper are computed using this extended set of countries. Nevertheless, the differences in average results for the extended set of countries and the country sample used in the 2005 are negligible.
5. This paper draws on a study which was originally commissioned by the European Commission to update the elasticities used to apply structural budget surveillance to EU member countries. The OECD report on EU elasticities was published in December 2014 (Price, Dang and Guillemette, 2014).
The social security contribution elasticities take account of employer contributions as well as employee contributions.
With respect to the elasticities of corporate income taxes and indirect taxes relative to their bases (profits and consumer spending respectively), the assumption that both have unit elasticities has been investigated empirically and the assumption of a unit elasticity for both taxes has been dropped in favour of regression-based elasticities. This contrasts with the Commission approach in which the indirect tax elasticities are assumed to be one.
The links between non-tax revenues and the output gap have been investigated, but no significant relationship to the cycle has been found.
In addition to re-estimating the links between unemployment–related transfers and the cycle, the relationship between other income related transfers (family benefits, housing benefits and in- work benefits) and the cycle has been examined. The inclusion of these transfers in the estimation process of the cyclically adjusted balance is a second source of difference between the OECD and the EC.
With respect to the regression analysis of tax revenue to tax base elasticities and of tax base to output gap elasticities, the econometric approach has continued to be based on individual country regressions, rather than a pooled cross-section approach, largely because the characteristics of tax systems are so different internationally. However, the regression approach has been refined and re-specified to distinguish between short-term and long-term elasticities.
3. The paper is organised as follows. The next section gives an overview of the methodology which is elaborated further in the Annex. The third section computes the elasticities of government revenues and the fourth section the elasticities for government non-interest spending. The final section presents a measure of the overall sensitivity of the budget balance, with new cyclically-adjusted budget estimates.
II. Conceptual and methodological issues
4. The cyclically adjusted balance (CAB) measures the underlying fiscal position, removing the effects of the business cycle from the budget, the process of adjustment being applied in disaggregated form to the principal categories of revenues and to unemployment-related government spending in levels, such that:
𝑏∗ = [∑𝑛𝑖=1𝑇𝑖∗− 𝐺∗ + 𝑋]/ 𝑌∗
(1) where: b* = cyclically adjusted fiscal balance as ratio of potential output; Y* = the level of potential output; Ti* = the ith category of cyclically adjusted tax; G* = cyclically adjusted government current primary expenditures (i.e. government spending excluding capital and interest spending), and X = capital and net interest spending and non-tax revenue.6
6. This disaggregation of government spending into current and other spending is an OECD convention; it is not an essential part of the method. As will be seen below, the main issue is to identify those components of spending which are responsive to the cycle.
5. The adjustment is based on the elasticities of the respective tax categories, and of unemployment related spending, with respect to the output gap, so that:
𝑏∗ = [ ∑ 𝑇𝑖
𝑛
𝑖=1
(𝑌∗⁄ )𝑌 𝜀𝑡𝑖.𝑦 − 𝐺(𝑌∗⁄ )𝑌 𝜀𝑔,𝑦+ 𝑋] /𝑌∗
(2) where: 𝑌∗⁄ = ratio of potential to the actual output (a measure of the output gap),T𝑌 i = actual revenue of the ith category of tax, G = actual government current primary expenditures,εti,y = the elasticity of the ith tax category with respect to the output gap and εg,y = the elasticity of current primary government expenditures with respect to the output gap.
6. The OECD/EC method for calculating the elasticities uses a two-stage approach, which identifies separately i) the elasticity of revenues and expenditures with respect to their base (Ɛt.tb), ii) the elasticity of bases with respect to the output gap (Ɛtb.y):
𝜀𝑡𝑖,𝑦 = 𝜀𝑡𝑖,𝑡𝑏𝑖 𝜀𝑡𝑏𝑖,𝑦 𝑎𝑛𝑑 𝜀𝑔,𝑦 = 𝜀𝑔,𝑢 𝜀𝑢,𝑦
(3) The tax revenue to tax base elasticities, (𝜀𝑡𝑖,𝑡𝑏𝑖 ) depend on the relevant tax codes, while the elasticities of the tax bases with respect to Y*/Y (𝜀𝑡𝑏𝑖,𝑦) are empirically estimated. Similarly, the government spending elasticities are composed of an elasticity of spending with respect to unemployment (𝜀𝑔,𝑢 ) and an elasticity of unemployment with respect to Y*/Y (𝜀𝑢,𝑦 ).
7. The 2005 model identifies four tax categories as being cyclically sensitive: direct taxes on households (personal income tax), social security contributions, corporate income tax and indirect taxes, the respective bases being taken as earnings (for income tax and contributions), the gross operating surplus for corporate income and consumption for indirect taxes. Only one category of spending is treated as cyclically sensitive – that relating to unemployment.
8. With respect to the calculation of the CAB, the OECD uses a disaggregated approach as per equation (2) : it first adjusts the individual tax and spending categories for the cycle, and then aggregates the resulting cyclically adjusted items, together with the non-tax revenue and capital and net interest spending, into a CAB. It is also possible to measure the responsiveness of the total budget balance-to-GDP ratio to the business cycle with a single number – the budgetary semi-elasticity. As opposed to an elasticity, which relates a percentage change of a tax (or spending) category in nominal terms to a percentage change in the output gap level, a semi-elasticity measures the absolute change of the budget balance-to-GDP ratio in response to a percentage change in GDP due to the business cycle. The budgetary semi-elasticity can be derived on a bottom-up basis from the individual tax and spending items, their output gap elasticities and relative weights.7 The cyclical adjustment process of the EC applies this semi- elasticity directly to the budget balance to derive a CAB.
Scope for enhancement
Further disaggregation
9. The 2005 model is based on the national accounts (SNA) definitions and is fairly comprehensive in its coverage of revenues (Table 1). However, the level of aggregation involved leads to a considerable
7. For more details, see Section V below and Mourre et al. (2013).
degree of simplification with respect to the taxes identified in OECD Revenue Statistics, some elements of which offer the potential for improvements.
The personal income tax (PIT) is related to wages and salaries in the 2005 model, whereas it is also levied on self-employment income, capital income, capital gains and (some) transfers, which may show a different cyclical behaviour from earnings. Realised capital gains are difficult to include in the cyclical adjustment process without specifying asset price cycles (Price and Dang, 2011). However, to the extent that capital income (interest, dividends etc.) is correlated with capital gains, some of the elasticity effects of asset prices movements are captured. Since some transfers are not taxed, the present study does not estimate separate elasticities for this income category.
The cyclical adjustment process equates PIT to 'direct taxes on households', which according to National Accounts definitions also includes taxes on immovable property and net wealth. Wealth taxes are applied by only a small number of OECD economies. Property taxes are a more widely used source of revenue and may be related to the output cycle via new house building, as well as to asset-price cycles via house prices.8 However, they only amount to 2.3% of total revenues on average (Table 1) so that the 2005 approach of not identifying separate elasticities for this tax category has been maintained.
Indirect taxes (IT) are levied on a number of expenditure bases, including intermediate goods and some elements of investment – residential building and renovation – which may be more cyclically sensitive than consumption. These tax bases should, in principle, be unbundled from consumption9. Similarly, indirect taxes also include taxes on financial transactions, which are likely to exhibit a different degree of cyclicality from consumption.10 However, the need to keep the updating exercise both uniform and tractable has prevented a move towards disaggregating indirect taxes, which continue to be related only to consumption.
There is a category of revenues coming under the National Accounts rubric of 'capital taxes' levied on inheritances and gifts which is excluded from the adjustment process. While these taxes may be related to the business cycle indirectly, via asset prices, their weight is very small.
Non-tax revenues have an important weight – about 20% of overall revenue on average - and the possibility that these may be cyclically related is examined here.
8. According to Price and Dang (2011), the relationship is not close, because of a lack of systematic revaluation in many countries.
9. They have been the origin of cyclical volatility in receipts not picked up in the existing adjustment method in the past.
10. OECD research shows that revenues in these categories are related to asset-price changes and not to GDP;
though in some countries asset prices can be partially correlated to the business cycle, overall, the relationship is weak. For further analysis of these issues see Price and Dang (2011).
Table 1. Categorisation of taxes and transfers and their bases
1. Shares are averages over the period 2001-2011 and unweighted averages of OECD member countries.
Source: OECD Revenue Statistics, OECD National Accounts Database, OECD Social Expenditures Database (SOCX).
Elasticity estimation processes
10. In the case of personal income taxes and social security contributions, the tax revenue to tax base elasticities are derived from the national tax codes, while in the case of corporate income taxes and indirect taxes the 2005 model assumes unit elasticity. Taxes with an elasticity of 1 – proportional taxes – do not affect the budget balance to GDP ratio, which means that, in effect, the principal drivers of cyclical variation in the budget balance to GDP ratio are personal income taxes and the social security contribution system. In the current analysis, the assumption that corporate income taxes and indirect taxes have unit elasticities was dropped in favour of empirically estimated elasticities. A priori, this allows for greater cyclical revenue sensitivity and greater cross-country variation in aggregate elasticities.
24.2 Earnings
Transfers
Self-employment income Income from capital
Capital gains 1120
Property taxes Property values 4100 2.3
Taxes on net wealth Asset values 4200 0.4
19.8
Employee contributions 7.4
Employer contributions 12.4
7.7
D51 Gross operating surplus and capital gains 1200 7.7
27.4
Tax on general consumption Personal consumption
of which VAT on new housing and repairs Housing expenditure
Taxes on specific goods and services Personal consumption and intermediate goods
Taxes on financial and capital transactions Asset transactions 4400 1.2
0.3
Estate, inheritance and gift taxes Asset values 4300 0.3
20.6
8.4
Unemployment-related spending Unemployment 2.0
Income related and family benefits Earnings 6.4
OECD weights
% in total revenues/expen
ditures1
Taxes on personal income D51 1110
21.5
3000 National Accounts classification
SNA classification
(SNA)
Base Revenue
Statistics (RS)
5.Capital taxes 6. Non-tax revenues
GOVERNMENT EXPENDITURES 7. Government transfers
D21 5000 26.2
GOVERNMENT REVENUES 1. Direct taxes on households
2. Social security contributions
3. Corporate income taxes 4. Indirect Taxes D59
D29 Wages and salaries
11. The regression modelling framework, which is described below, now also allows for the calculation of short-run and long-run tax revenue to tax base elasticities, which can diverge because of collection lags or compositional changes within the tax base. This specification would seem well adapted to account for the actual cyclical behaviour of indirect and corporate income taxes - the short-term elasticity capturing temporary movements due to cyclical shocks.11 Reflecting institutional and behavioural differences, it is to be expected that the short-run elasticities will show greater international divergence than long-run elasticities. However, elasticities estimated in this way are also likely to be dependent on the time period involved and this poses problems for the cyclical adjustment process going forward, for instance where indirect taxes are subject to collection problems.
III. Revenue elasticities
1. Personal income taxes and social security contributions Tax and social security contribution elasticities with respect to earnings
12. While in the 2005 model the elasticities of income tax referred to wage income, these were equated, in practice, with the elasticities of income tax relative to all incomes. Indeed, earnings (defined as wages and salaries) are the largest part of the PIT base and account on average for around two-thirds of the base. Earnings also constitute the base for social security contributions, which means that, on average for OECD economies, around two-fifths of general government revenues are based on earnings, making this tax base one of the most important drivers of cyclicality.
13. The elasticities of PIT revenues relative to earnings are derived from average earnings data, which relate per capita income tax paid to incomes along a distribution scale measured in multiples of average earnings. For individuals/households with identical characteristics as to marriage status and children, the average and marginal taxes can be calculated from the relevant tax codes at each point along the earnings schedule (see Annex Part I).12 The aggregate average and marginal tax rates are then calculated by an income-weighting process, to provide the aggregate elasticity of tax relative to earnings εt,ye :
𝜀𝑡,𝑦𝑒 =𝑀𝑅
𝐴𝑅 = ∑ 𝜔𝑦𝑒,𝑖𝑚𝑟𝑖/∑ 𝜔𝑦𝑒,𝑖𝑎𝑟𝑖
(4) where ye,i = the weight of percentile earnings-level i in total earnings expressed in currency units earned, mri = marginal income tax rate (social security contribution rate) at point i on the earnings distribution and ari = average income tax rate (social security contribution rate) at point i on the earnings distribution. MR and AR are the weighted marginal and average rates of tax, respectively. Both the definition of the 'representative' taxpayer and the income-weighting process used to generate the aggregate marginal and average rates are critical to the calculation process.
14. In the 2005 model, a log-normal distribution was fitted along an income distribution scale from zero to three times average earnings, to arrive at an aggregate elasticity for a representative average production worker, defined as a full-time two-earner married couple with two children.13 The new calculations are based on an income scale which now covers zero to up to eight times average earnings and
11. See Belinga et al. (2014) for a similar approach.
12. The data refer to gross earnings of full-time workers by earnings percentiles in national currency units. The earnings by deciles are available from the OECD Labour Market Statistics.
13. The secondary earner is on 50 per cent of the average production worker earnings.
the parameters governing the log-normal distribution have been based on actual income distribution data.14 To reduce the sensitivity of the results to the representative family type chosen, three household types have been averaged to produce the estimated PIT and employee social security contribution elasticities, rather than relying on a single family type.15 The categories are i) single persons; ii) married couple with a single earner and no children; and iii) a married couple with two children, the second earner on two-thirds of average income. Data are not available as to the share of taxation paid by each family category, so they cannot be weighted to compute an average. An arithmetical average of the three types is used, as it is likely to be a more reliable estimate of the aggregate elasticity than applying the elasticity of a single component.
15. The elasticities of PIT revenue to earnings so calculated and the effects of applying the new parameterisation under the 2005 methodology are shown in Table 2 (see col. 2). The elasticities vary between 1¼ and 2¾. A prominent feature of the results is thus the rather wide dispersion of elasticities, around an OECD average of 1¾. These differences are explicable in terms of the varying tax structures in operation, ranging from a fairly flat and uniform structure (Denmark, Latvia) to one where allowances and exemptions determine that tax only starts to be paid well up the income scale. Higher thresholds tend to push up the aggregate tax elasticity for a given tax schedule (see discussion in Annex Part I).
16. Table 3 compares the new PIT elasticity estimates with those of the 2005 exercise, decomposing the changes into statutory and methodological changes.16 The effect of statutory changes is reported in OECD (2012a)17, while the impact of the methodology can be gauged by applying the 2005-model income distribution and representative-agent parameters to the 2010 data set (see Annex Part I). The overall impact of method and rate changes has been to increase the PIT-to-earnings elasticities for a half of OECD economies and reduce it for the other half. The revisions are in the range of + or – 0.6 (with the exception of Slovakia), though mostly of the order of + or – 0.2. The causes of the revisions are complex and discussed in greater detail in the Annex, but certain factors can be quantitatively identified:
Statutory changes have had mixed effects.18 In about a half of the economies covered by the 2005 study, the elasticity has declined, because of reduced thresholds or reduced tax progression. For the other half higher thresholds and/or more progressive rate structures (the former being more important) have pushed the elasticity up, though the effect is much more marginal.
Methodological change: The broader definition of the representative family has had the effect of reducing elasticities, on average, because families with two children tend to have higher thresholds, which reduces the average rate of tax relative to the marginal and results in higher elasticities. Including families with no children hence reduces the aggregate elasticities. Changes in the income-weighting system (extending the analysis to higher income earners) have also had the effect of reducing income tax elasticities, on average, insofar as higher income earners face a lower tax elasticity and these are given a greater weight (see Annex Part I discussion). This would appear to be the case for Germany, Spain, Italy and Sweden, for example.
14. These income distribution data are available from the Distribution and Poverty data set which gives income and tax data by population decile; the standard deviation of income derived from this distribution is used to calculate the lognormal income distribution applied to the Tax/benefits data set (see Annex).
15. An exception is Italy, where taking account of the income tax structure, the average of two family types has been used: i.e. families with 0 and 2 children.
16. In some cases (Greece, Slovakia, for example) the 2005 exercise based the elasticity estimate on conventional assumptions rather than calculation.
17. Tax schedules are described for 2003 (the year used in the 2005 model calculations) and 2010, allowing the effects of statutory changes to be calculated between those two years.
18. Statutory effects have been estimated directly from the 2003 and 2010 income tax schedules given in OECD (2012a).
Table 2. Tax to tax base elasticities of income and social security contributions.
2005 and new methodology
Note: This table compares tax-to-tax base elasticities between 2005 and new methodology. First, personal income tax elasticities now take account of self-employment and capital incomes in addition to earnings. Second, employers’ social security contributions are also taken on board in the calculation of social security contributions (SSC) elasticities. Column 1 shows the PIT elasticities as derived in 2005. Compared to column 2, the differences reflect the changes in the tax system while applying the same methodology. However, compared to column 8, the differences reflect in addition the inclusion of other personal income components. Detailed calculations are shown in Table 4. Total revisions are indicated in column 9 and the sources of revisions are explained in Table 3.
Column 3 refers to the 2005 SSC elasticities and column 4 to the updates reflecting recent social security rates on the employees’ side. Compared to column 6, the difference reflects the employers’ contributions in the SSC elasticities.
OECD averages are unweighted averages.
1. The published tax data have been adjusted to correspond to 2014 PIT tax schedule.
2. The elasticity is an average of married couple with 0 and 2 children, to take into account of the particular tax structure.
Source: OECD calculations, Taxing Wages (OECD 2012a), Girouard and André (2005).
2005
estimates New estimates 2005 estimates (employee)
New estimates (employee)
New estimates (employer)
New estimates
(total) New estimates New
estimates Revision New estimates
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
Australia 1.50 1.62 0.00 0.00 0.00 0.00 1.62 1.60 0.10 1.60
Austria 2.20 2.00 1.00 0.85 0.99 0.92 1.25 1.97 -0.23 1.34
Belgium 1.60 1.63 1.10 1.30 1.00 1.15 1.34 1.62 0.02 1.36
Canada 1.60 2.06 0.80 0.70 0.71 0.71 1.42 2.04 0.44 1.65
Chile .. 2.76 .. 1.00 0.00 1.00 1.47 2.16 .. 1.14
Czech Republic 1.70 2.24 1.10 0.98 0.99 0.99 1.11 2.23 0.53 1.24
Denmark 1.40 1.44 1.00 0.70 0.00 0.70 1.23 1.43 0.03 1.38
Estonia .. 1.46 .. 1.00 1.40 1.36 1.39 1.46 .. 1.39
Finland 1.50 1.50 1.00 1.02 1.00 1.00 1.22 1.48 -0.02 1.25
France¹ 1.70 1.73 1.10 0.91 0.96 0.95 1.12 1.68 -0.02 1.20
Germany 2.30 1.90 0.80 0.76 0.97 0.86 1.13 1.88 -0.42 1.22
Greece 2.00 2.30 0.90 0.80 0.86 0.84 1.10 2.21 0.21 1.26
Hungary 2.43 1.84 0.90 1.00 0.99 0.99 1.21 1.80 -0.63 1.23
Iceland 1.44 1.73 1.00 1.25 1.00 1.05 1.42 1.72 0.28 1.58
Ireland 2.10 2.11 1.30 1.49 1.41 1.44 1.83 2.04 -0.06 1.82
Israel .. 1.94 .. 1.37 1.16 1.23 1.54 1.83 .. 1.51
Italy² 2.00 1.84 1.00 1.00 0.96 0.97 1.23 1.85 -0.15 1.39
Japan 2.00 1.88 0.90 0.89 0.88 0.88 1.13 1.87 -0.13 1.19
Korea 2.34 2.36 0.90 0.83 0.88 0.86 1.23 2.24 -0.10 1.42
Latvia .. 1.29 .. 1.00 1.00 1.00 1.10 1.31 .. 1.12
Luxembourg 2.50 2.28 1.30 0.90 0.93 0.91 1.22 2.24 -0.26 1.39
Mexico .. 2.22 .. 1.11 1.06 1.07 1.32 2.08 .. 1.72
Netherlands 2.40 2.15 0.80 0.84 0.71 0.80 1.35 2.00 -0.40 1.21
New Zealand 1.30 1.38 0.00 0.00 0.00 0.00 1.38 1.35 0.05 1.35
Norway 1.50 1.53 1.10 1.04 1.00 1.02 1.26 1.53 0.03 1.29
Poland 1.40 1.96 1.00 0.96 0.98 0.97 1.08 1.93 0.53 1.23
Portugal 1.70 2.22 1.00 1.00 1.00 1.00 1.24 2.15 0.45 1.40
Slovak Republic 1.00 2.47 1.00 0.97 0.98 0.98 1.10 2.43 1.43 1.26
Slovenia .. 2.15 .. 1.00 1.00 1.00 1.21 2.14 .. 1.34
Spain 2.10 1.93 0.80 0.88 0.82 0.83 1.13 1.88 -0.22 1.23
Sweden 1.30 1.45 1.00 0.69 1.00 0.95 1.16 1.42 0.12 1.26
Switzerland 1.84 1.92 1.20 0.79 0.87 0.83 1.05 1.87 0.03 1.47
Turkey .. 1.53 .. 0.98 0.97 0.97 1.13 1.51 .. 1.20
United Kingdom 1.70 1.50 1.30 0.97 1.33 1.20 1.35 1.49 -0.21 1.37
United States 1.90 1.65 0.90 0.93 0.79 0.85 1.20 1.64 -0.26 1.32
OECD 1.80 1.88 0.90 0.91 0.87 0.92 1.27 1.83 0.04 1.35
Income tax
Income tax and social security contributions
Social security contributions Income Tax
Tax elasticities relative to earnings Tax elasticities relative to total personal income - SNA D51
Income tax and social security
contributions
Table 3. Source of revisions of tax to tax base elasticities for personal income and social security contributions
Note: OECD averages are unweighted averages.
1. Includes the impact of new income distribution assumptions (see Annex) and other unidentified data-related factors.
2. Difference between column 6 and column 4 in Table 2.
Source: OECD calculations.
Taxes on other components of personal income
17. The 2005 model takes the personal income tax base solely as earnings, though, as noted, these are only one component of the base, which also includes self-employment income and capital income (Table 1). The tax revenues to tax base elasticities applying to these other components are likely to differ from that applying to earnings, but, more importantly, the relationship of the bases to the output gap is
Policy Policy and
methodology Methodogy
Statutory rate
changes Income distribution1 Representative
type Inclusion of non-
earnings income Employee contributions
Inclusion of employer's contributions2
[1] [2] [3] [4] [5] [6] [7]
Australia 0.10 0.07 0.04 0.00 -0.02 0.00 0.00
Austria -0.23 -0.05 -0.08 -0.07 -0.03 -0.15 0.07
Belgium 0.02 -0.03 0.03 0.03 -0.02 0.20 -0.15
Canada 0.44 0.19 0.31 -0.03 -0.03 -0.10 0.00
Chile .. .. .. .. .. .. ..
Czech Republic 0.53 -0.35 0.95 -0.06 -0.01 -0.12 0.00
Denmark 0.03 -0.03 0.05 0.02 -0.01 -0.30 0.00
Estonia .. 0.10 .. .. .. .. ..
Finland -0.02 -0.02 0.07 -0.05 -0.02 0.02 -0.01
France -0.02 -0.29 0.32 0.00 -0.05 -0.19 0.04
Germany -0.42 0.10 -0.51 0.01 -0.02 -0.04 0.10
Greece 0.21 -0.22 0.71 -0.19 -0.09 -0.10 0.04
Hungary -0.63 -0.55 0.02 -0.07 -0.03 0.10 -0.01
Iceland 0.28 0.14 0.00 0.15 -0.01 0.25 -0.20
Ireland -0.06 0.16 -0.01 -0.13 -0.07 0.19 -0.06
Israel .. .. .. .. .. .. ..
Italy -0.15 0.05 -0.18 -0.03 0.01 0.00 -0.03
Japan -0.13 -0.07 0.07 -0.12 -0.01 -0.01 0.00
Korea -0.10 -0.63 0.85 -0.21 -0.12 -0.07 0.03
Latvia .. .. .. .. .. .. ..
Luxembourg -0.26 -0.09 0.01 -0.15 -0.04 -0.40 0.01
Mexico .. .. .. .. .. .. ..
Netherlands -0.40 -0.24 0.19 -0.20 -0.16 0.04 -0.04
New Zealand 0.05 0.13 -0.05 0.01 -0.03 0.00 0.00
Norway 0.03 0.03 -0.10 0.10 0.00 -0.06 -0.03
Poland 0.53 -0.43 1.75 -0.76 -0.03 -0.04 0.01
Portugal 0.45 0.00 1.03 -0.52 -0.06 0.00 0.00
Slovak Republic 1.43 0.21 1.20 0.06 -0.03 -0.03 0.00
Slovenia .. .. .. .. .. .. ..
Spain -0.22 0.06 -0.17 -0.06 -0.05 0.08 -0.05
Sweden 0.12 0.44 -0.26 -0.02 -0.04 -0.31 0.26
Switzerland 0.03 0.02 0.24 -0.17 -0.05 -0.41 0.04
Turkey .. .. .. .. .. .. ..
United Kingdom -0.21 0.03 -0.25 0.02 0.00 -0.33 0.23
United States -0.26 0.04 -0.19 -0.10 -0.01 0.03 -0.08
OECD 0.04 -0.04 0.22 -0.09 -0.04 -0.06 0.01
Total revision
Revision due to: Revision due to:
Personal income tax elasticity Social security contribution elasticity
Methodology
likely to be quite different. In particular, for some countries, treating capital income as equivalent to earnings for cyclical adjustment purposes may have been a source of some error, because of the heightened cyclicality of dividends and capital gains. The estimation of aggregate PIT revenue elasticities with respect to total personal income and of total personal income with respect to the output gap thus requires the incorporation of the elasticities of capital and self-employment income into the cyclical adjustment process.
18. As is the case for earnings-related taxes, a two-stage approach has been adopted for the estimation of tax/output gap elasticities relating to non-earnings income components, as per equation 3.
The tax/tax base elasticities applying to self-employment and capital income have been computed from cross-section income distribution data (the Distribution and Poverty data set). Separate tax/income elasticities can be inferred by applying the respective aggregate income weights to each average-income category, as in equation 4, except that the data exist only by decile and only in respect of the combined total of PIT plus employee social security contributions (for more details, see Annex). In general, taxes on capital income and self-employment income have lower elasticities with respect to their respective bases compared with those applying to earnings, driven by income distribution differences (Table 4, cols. 2-4).
For OECD countries on average, the PIT/tax base elasticity remains unchanged compared to 2005, at 1.8 (Table 2 col. 8 and Table 4 col. 1).
Social security contributions
19. Employee social security contribution data are available in exactly the same form as personal income tax data in the Tax/benefits data set and an identical procedure has been applied to arrive at an elasticity of social contributions with respect to earnings, as in the previous model (Table 2, col. 4). In the 2005 model, the elasticities applying to employers' contributions, which are not covered by the average earnings data set, are assumed to be equal to those applying to employees’ contributions. In fact, however, rates of employee and employer contributions usually differ and here the aggregate average and marginal rates of employers’ contributions are calculated independently, and employers' contributions added to employee contributions at each point in the income distribution, i, based on the actual operational parameters of the contributions system. This allows the calculation of a total contributions/earnings elasticity according to equation 4 (Table 2, col. 6).
20. In aggregate, both employee and employer social security contributions increase less than proportionally to the per capita earnings base, since they are usually specified at a flat rate up to a statutory ceiling. Five OECD countries have unit elasticity, a quarter have progressive contributions and the remainder has regressive contributions. The reduced progressivity built into the system of social security contributions thus offsets to some extent the progressivity of the PIT, which ensures that the combined PIT and social security contributions elasticity is lower than the PIT elasticity in almost every case. The combined PIT and social security contribution elasticities range from 1.05 (Switzerland) to 1.8 (Ireland), with an OECD average of around 1¼ (Table 2, col. 7).
21. The sources of the revisions to social security contribution/earnings elasticities due to methodological and policy adjustments to employee contributions and to the inclusion of employer contributions are given in Table 3 (cols. 6 and 7). In general, the latter adjustments are small (col. 7), Iceland, Sweden and the United Kingdom being exceptions. Revisions to the employee contributions elasticity due to policy and methodology are somewhat larger (col. 6) and negative in several cases, and more marked in Switzerland, Luxembourg, the United Kingdom, Sweden, and Denmark, while the Belgian, Icelandic and Irish elasticities are relatively higher among those being revised up.
Table 4. Computation of tax to tax base elasticity of personal income and its components
Note: OECD averages are unweighted averages.
1. See Table 2 column 8.
2. See Table 2 column 2.
Source: OECD calculations, see Methodological and Statistical Annex Table A1.2 for detailed explanations.
Earnings2 Self-employment
income Capital income Earnings Self-employment incomes
Capital income
[1] [2] [3] [4] [5] [6] [7]
Australia 1.60 1.62 1.47 1.51 0.74 0.06 0.10
Austria 1.97 2.00 1.85 1.70 0.61 0.10 0.04
Belgium 1.62 1.63 1.39 1.69 0.68 0.07 0.04
Canada 2.04 2.06 1.60 2.12 0.70 0.07 0.12
Chile 2.16 2.76 1.31 1.28 0.57 0.27 0.13
Czech Republic 2.23 2.24 2.24 1.77 0.63 0.14 0.02
Denmark 1.43 1.44 1.38 1.39 0.71 0.05 0.07
Estonia 1.46 1.46 1.45 1.46 0.76 0.02 0.01
Finland 1.48 1.50 1.43 1.32 0.65 0.06 0.05
France 1.68 1.73 1.69 1.38 0.58 0.06 0.10
Germany 1.88 1.90 1.87 1.74 0.57 0.17 0.06
Greece 2.21 2.30 2.14 1.59 0.49 0.22 0.05
Hungary 1.80 1.84 1.74 1.50 0.47 0.07 0.04
Iceland 1.72 1.73 1.44 1.67 0.74 0.02 0.07
Ireland 2.04 2.11 1.61 1.81 0.60 0.09 0.01
Israel 1.83 1.94 1.43 1.61 0.67 0.13 0.09
Italy 1.85 1.84 1.89 1.75 0.49 0.21 0.04
Japan 1.87 1.88 1.78 1.80 0.69 0.05 0.06
Korea 2.24 2.36 1.97 1.98 0.66 0.23 0.06
Latvia 1.31 1.29 1.24 1.60 0.65 0.10 0.06
Luxembourg 2.24 2.28 1.92 1.86 0.67 0.05 0.03
Mexico 2.08 2.22 1.45 1.64 0.73 0.12 0.06
Netherlands 2.00 2.15 1.84 1.20 0.67 0.08 0.11
New Zealand 1.35 1.38 1.30 1.23 0.65 0.10 0.12
Norway 1.53 1.53 1.54 1.52 0.67 0.05 0.08
Poland 1.93 1.96 1.84 1.51 0.64 0.12 0.02
Portugal 2.15 2.22 1.73 1.91 0.66 0.09 0.02
Slovak Republic 2.43 2.47 2.20 1.93 0.68 0.09 0.01
Slovenia 2.14 2.15 2.19 1.64 0.71 0.05 0.02
Spain 1.88 1.93 1.48 1.83 0.67 0.08 0.02
Sweden 1.42 1.45 1.21 1.17 0.68 0.03 0.08
Switzerland 1.87 1.92 1.46 1.80 0.72 0.07 0.10
Turkey 1.51 1.53 1.45 1.51 0.49 0.21 0.11
United Kingdom 1.49 1.50 1.49 1.48 0.67 0.08 0.09
United States 1.64 1.65 1.50 1.62 0.77 0.05 0.08
OECD 1.83 1.88 1.64 1.61 0.65 0.10 0.06
Personal income tax / tax base elasticity1
Direct tax / tax base elasticities of income
components Income component tax weights
